The Optimized Comparison of the Gray Model Improved by Posterior-Error-Test and SVM Modified by Markov Residual Error in the Long-medium Power Load Forecast

Generally, the long-medium power load forecasting sequence has small sample, stochastic growth and nonlinear wave characteristics. Gray and SVM model could reflect the relationship between growing characters-tics and nonlinear characteristics to the series effectively and make fitting calculation. The paper modifies the proposed gray model through posterior-error-test and compares the predictive value of power load when the evaluation result is best with the optimal result forecast by SVM that is modified by Markov residual. Then we can find which model is the better. As result, we can see that Markov could well reflect randomness that produced by the system involve with many complex factors. A forecast model based on SVM algorithm is established, the series of historical load variables is rolling forecasted. It is proved that the presented forecast method is superior obviously to traditional methods through empirical study, and it can be used generally.